The state of the art of business application monitoring

Since the 1980s, Windows applications have been continuously evolving, with a technological disruption every 20 years. The first native applications have now made way for published or stremmed applications, with Citrix’s arrival on the scene in the 2000s as well as Virtual Desktop infrastructure (VDI).

The Desktop as a Service (DaaS) revolution

By 2017, the new generation of streamed applications, including DaaS, was born. The emblematic provider was Amazon, with its AppStream V2 service. Positioned among its competitors was Citrix, with its Horizon offer.
Starting out with small businesses and SMEs, this type of offer largely attracts large organizations, particularly in the banking, health, and online education sectors. The market has strong potential, according to Gartner’s findings:

“The worldwide public cloud services market is forecast to grow 6.3% in 2020 to total $257.9 billion […]. DaaS is expected to have the most significant growth, increasing 95.4% to $1.2 billion.” Gartner, Inc.

The new deployment models for these new applications

There are three deployment types, which can be set up in accordance with the needs, selection criteria, and circumstances in your enterprise:

• internal VDI: You deploy your own infrastructure and applications; you are therefore entirely responsible for your environment.
• managed VDI: You call on a provider to handle infrastructure deployment, and you keep control over your applications.
• DaaS: Your provider is fully responsible for all services, including applications.

Added to these are three quality control models:

• internal VDI: You are free to choose any type of quality measurement system.
• managed VDI: The choice of monitoring tools is negotiated with your provider in light of common objectives.
• DaaS: You do not have control over the technical infrastructure. The provider must supply performance KPIs based on its own tools. Client monitoring with your own tools is still possible, but must be non-intrusive.

DaaS and VDI: how do they work?

Historically, applications were installed on Windows client workstations. Eventually, since the Citrix years, people now have been able to choose between thick client and thin client. The DaaS revolution thus brings a 100% web rationale.
How does it work? The session is managed remotely via a gateway (streaming gateway or VDI gateway, depending on the technology). It is in the VDI provider’s infrastructure that the applications of the enterprise or customer are published.

Types of APM solutions to use

First of all, it is important to note that Real-User Monitoring (RUM) is not the right solution to measure the performance of applications provided with VDI and DaaS. RUM is, in fact, a technology for the world of the web based on JavaScript code. It is therefore unable to efficiently measure Windows infrastructures.

Now let’s compare 4 types of APM solution:

• Vendor APM: Each DaaS or VDI provider usually offers a first level of monitoring based on tracking usage data.
• Server APM: These are APM solutions which require code to be supplied on the servers under monitoring.
• Network APM: These solutions operate by analyzing network traffic between the user and the application server under test.
• Robot APM: These non-intrusive solutions are based on “robots” automatically replaying user journeys.

Vendor APM, Server APM, and Network APM bring you a technical view of measurement, including analysis of client-side or server-side traffic, components and network latency, and all instances on the DaaS.
When the application is on the DaaS, Server APM cannot be used because there is no access to the provider’s servers.
Only the robot can supply precise user experience metrics. The robot provides indicators about page loading, clicks, and executions and interactions. These are real indicators from the end-user point of view.

Vendor APM, Network APM, Server APM or Robot APM?

These different APM solutions work together. The portion of the audience measured is an important factor. Vendor APM, Network APM and Server APM measure 100% of the audience, 100% of instances and 100% of sessions.
As for the robot, it confines itself to key user journeys (without any limit on scenario executions), aiming for a ‘lifelike’ approach.
The strength of Server APM is its full stack analysis which offers a complete, rich view if the results are skillfully interpreted.
Lastly, the robot is very appropriate for its client view for video and replay. It can highlight what’s wrong, but it needs to be paired with an internal solution to gain access to root causes.

The 6 main needs to consider in choosing a Windows monitoring solution:

• Windows efficiency: The solution must easily be able to handle the Windows environment (with key functions provided for).
• Web efficiency: It must be at ease with web, not just Windows.
• Cloud, intranet, and data center environments: The solution must be effective in any environment where there are client measurements to conduct.
• Windows, VDI and DaaS: The solution must be able to handle these 3 types of deployment in accordance with the infrastructure.
• Incident resolution: The solution must be able to measure, process SLAs, and help resolve incidents.
• User friendliness: It must be straightforward to use by internal teams.

Server APM and Network APM solutions are meant to be used by people with IT specialist profiles. Only Robot APM is easily accessible to business profiles. The choice must be made in line with the enterprise’s objectives and needs, but should also take into consideration the intended audience of your APM solution.

Moreover, with the DevOps revolution, APM is no longer all by itself! Solutions must additionally provide for:

• test automation
• load testing
• continuous monitoring